Interesting things in the world of electronics served as easy to digest articles with a touch of wit.

News articles of GaN chips float around stating the usefulness and importance. Have you ever wondered what they refer to or how they are different from silicon? Let us find out.

HEMT is a type of field effect transistor (FET) where the conduction channel is formed by a very dense layer of electrons called a two-dimensional electron gas (2DEG). That was a lot of big words in a small amount of time; field effect means that the transistor operation is voltage controlled and applying a voltage at the gate terminal will change the output characteristics of the device. A two-dimensional electron gas refers to an area with very high number of electrons forming a layer in a material. The layer has almost no thickness causing it to be called two dimensional. These devices are made of AlGaN/GaN, AlGaAs/GaAs and other III-V compound heterojunctions.

What is a heterojunction?

Two different materials forming an interface with each other is called a heterojunction. For example a P-N junction is a homojunction because the doping is different but the parent material is the same, i.e. silicon. A heterojunction is AlGaAs/GaAs junction where these two materials possess different bandgaps, lattice parameters, and other dissimilar properties.

The quality of a heterojunction is directly related to the performance of a HEMT. It is the most crucial part of the device during fabrication as increased defect density at the interface can reduce 2DEG confinement leading to higher electron scattering which will in turn raise the magnitude of on-resistance. Contamination introduces trap states leading to noise and crystal defects like dislocations, and dangling bonds can introduce threshold and breakdown voltage instability.

Why does HEMT possess high mobility?

Electron mobility (μ) (units: cm²/V.s) describes the ease with which electrons drift in the material. It is expressed as the average particle drift velocity per unit electric field. Mobility is a property of the semiconductor material used to make the device; it is affected by scattering events taking place in the bulk of the material. The main types of scattering events taking place in HEMT are:

Phonon Scattering

In quantum mechanics, many physical calculations are done in the form of discrete values or packets. This is where the term quantized form comes from. A photon is a discrete form of electromagnetic radiation or light particles and similarly a phonon is a discrete form of lattice vibrations in a solid. Phonons are used to study heat transport and its effect on material properties. This type of scattering has two types:

• Acoustic Phonon Scattering: Acoustic phonons are lattice vibration modes in which neighbouring atoms move in-phase, i.e., move nearly together at long wavelengths. They share similarities to sound waves traveling in a solid medium. At high temperatures, lattice vibrations can distort the crystal structure causing electron scattering.

μ_acoustic ∝ T^-1.5

Where ‘T’ is temperature and ‘μ_acoustic‘ is acoustic mobility.

• Optical Phonon Scattering: They are the opposite of acoustic phonons, the neighbouring atoms move out of phase. These vibrations interact strongly with electromagnetic radiation. In polar semiconductors like GaN and GaAs, these oscillations create oscillating electric dipoles. Electrons can absorb or emit photons leading to large change in kinetic energy. Since, the interaction is stronger, mobility can collapse rapidly once optical phonon emission becomes active.

$\mu_{op} \propto e^{\hbar\omega_{op}/kT}$

where ‘μ_op‘ is optical mobility, ‘h’ is Planck’s constant, ‘k’ is Boltzmann’s constant, ‘T’ is temperature, and ‘ω_op‘ is optical frequency.

Impurity Scattering

This scattering is caused by foreign particles or dopants. It is one of the defining scattering mechanisms in a device due to it being dependant on the material purity and doping concentration. Electrons are deflected from their trajectory due to irregularities in an otherwise periodic lattice. Two types exist:

• Ionized Impurity Scattering: When dopant atoms are activated, i.e., they start to donate or accept electrons, they become electrically charged. Electric charge creates electrostatic forces or long-range Coulomb forces which can attract or repel electrons changing their path.

The relation between impurity scattering and mobility is given by the Brooks-Herring Model:

The important relation we need to focus on is:

• Neutral Impurity Scattering: This is caused by electrically inactive impurities like residual contamination, un-ionized dopants, vacancies, etc. These impurities change crystal properties but are very weak in scattering ability and are very localized in their effect.

Defect Scattering

These defects arise from imperfections in the crystal lattice during fabrication. GaN HEMT can suffer from a lot of defects due to lack of large, high quality bulk GaN substrates. Structural irregularities in the crystal cause electron deflection, like a diversion sign on the highway.

Interface Scattering

The interface created by the heterojunction will have defects due to the difference in structural properties of the two constituents. The roughness at the interface can change electron direction due to local energy changes. Trap states and dangling bonds can stop electrons from reaching the conduction layer lowering channel density.

Alloy Disorder Scattering

The barrier layer is made of a ternary compound, i.e., AlGaAs or AlGaN. In ternary compounds, some lattice sites get substituted with another element. In this case, aluminum replaces gallium. At an atomic level, the film does not have uniform composition, some lattice sites have a higher concentration of Al than Ga. Atoms do not form a perfect periodic arrangement. Al and Ga have different properties like electronegativities and bond potentials. These differences bring changes in the local electronic environment, deviating it from our ideal scenario.

Carrier-Carrier Scattering

The charge carriers themselves possess a polarity which causes attraction or repulsion between particles. In devices with majority charge carriers as electrons, the most important scattering in this category is electron-electron scattering. It does not have nearly high enough magnitude of scattering since momentum is mostly preserved and energy is not lost to surroundings, it is redistributed among other electrons. This type of scattering plays an important role at high voltages where breakdown can occur.

Why did I explain scattering in such detail?

Coming back to the original question, HEMT possesses high mobility because electrons do not flow in the bulk of the structure but are confined to a very thin region. This separation allows suppression of ionized impurity defects and freedom from most bulk crystal defects.

Table 1: Representative data of bulk GaN and AlGaN/GaN HEMT. [Reference: Look and Sizelove (2001)]

Scattering Mechanism	Bulk Mobility Limit (cm²/V·s)	HEMT Mobility Limit (cm²/V·s)	Improvement Factor
Ionized Impurity (300K)	1,800	30,000	16.7×
Neutral Impurity (300K)	5,000	15,000	3×
Defect/ Dislocation	4,000	12,000	3×
Interface Roughness	N/A	8,000	N/A
Interface Charge	N/A	10,000	N/A
Alloy Disorder	N/A	12,000	N/A
Phonons (300 K)	1,500	3,200	2.1×

Total mobility is calculated using Matthiesen’s Rule:

As you can observe in the above figures, the major scattering event hampering electron mobility is scattering caused by ionized impurities. The new scattering events introduced due to heterojunction formation do not have nearly the same impact as ionized impurity scattering.

Where does high mobility help in field applications?

A higher mobility value can be injected in basic semiconductor formulae to gain an understanding about the advantages:

Higher current capability: $J=q n \mu E$

where J is current density, q is electron charge, n is electron density, μ is mobility and E is electric field.

Higher velocity of carriers: $v_d=\mu E$

where v_d is drift velocity.

Lower on-resistance: $R_{on}\propto\frac{1}{\mu}$

Most important use is higher switching frequency: $f_T\approx\frac{v_{sat}}{2\pi L_g}$

where f_T is current-gain cutoff frequency, defined as the frequency at which the transistor current gain becomes unity, v_sat is saturation velocity, and L_g is gate length.

HEMT Stack/Structure

A basic structure of a HEMT is given in the figure above. Layers are grown using metal organic chemical vapour deposition (MOCVD). The different layers all have crucial roles to play in its efficient working. Let us go from bottom to top:

Substrate: It forms the base for depositing epitaxial layers to form HEMT.

Nucleation Layer: It is also called transition layer. It bridges lattice and thermal constant differences and buries impurities at the surface of substrate otherwise, strain arises favouring the creation of cracks. To prevent this cracking, intermediate layers are introduced between the substrate and the channel layer to compensate for the significant lattice mismatch. This layer is not required if the channel layer and the substrate are identical, for example GaAs layer on GaAs substrate.

Buffer Layer: It is a thick layer consisting of a few microns. It helps in relaxing the film as a thick film will have minimal compressive stress. It counters stress to minimize wafer bowing, manages stress so electrical and mechanical properties remain intact. It also prevents current leakage into nucleation layer. The maximum voltage a transistor can sustain depends mainly on the buffer layer. For greater voltage a tall buffer stack must be deposited.

Channel Layer: Conduction channel is formed within this layer due to electrons accumulating due to the triangular quantum well.

Barrier Layer: It confines the conduction channel to the interface between barrier and channel layer. A layer with higher bandgap and lesser electron affinity is grown. Due to the conduction band offset a potential quantum well will be created underneath the hetero-interface, trapping the electrons inside, and consequently creating the channel. The depth of the well is associated with the difference of electron affinity. Bandgap engineering is the term used to create conduction band offset. To increase 2DEG density, this layer is doped with donors (n-type dopants).

Electrodes: Electrodes, named the source and the drain, are placed on highly doped semiconductor to achieve ohmic contacts. However, the third electrode, named the gate, is placed on non-heavily doped semiconductor to generate the Schottky barrier. The Schottky gate controls the carrier concentration in the channel layer below the interface. As the gate voltage decreases, the carrier concentration below the gate electrode decreases. The gate bias required to pinch-off the channel is called the threshold voltage (V_th). Below V_th, channel becomes depleted from carriers and, thus, no useful current can flow between the drain and source.

How is the conduction region formed?

The conduction band energy refers to the range of electron energies within a solid where electrons are free to move and contribute to electrical conduction. Electrons in the conduction band are not bound to any specific atom and can move freely under the influence of an electric field. This movement of electrons constitutes an electric current. The density of states in the conduction band, which describes the number of available electron states at each energy level, also influences how many electrons can occupy the conduction band and contribute to conductivity.

The barrier layer needs to be doped with donors to allow correct 2DEG formation. AlGaN/GaN has an intrinsic property of polarization of charges allowing a lower amount of modulation doping or none at all. This doping to create 2DEG in HEMT is called modulation doping.

The fermi energy level is the level where probability of finding electrons is ½. The formula for determining probability is:

f(e) = [1-e^(E-Ef)/kT]^-1

Where E_F is the Fermi Energy level. The Fermi level is usually present in the forbidden energy gap. No electrons can be present there because there are no states. States are like houses for electrons. Without a state, no electron can take up residence. When the conduction band goes under the fermi level, a huge condition is fulfilled, states and a huge probability are present together. Electrons will occupy those states and the conduction channel, i.e. 2DEG will be formed. The states exist slightly below the interface in the channel layer, this is why, the 2DEG shown in figure (4) is around a few nanometers beneath the interface. Band alignment is in the form of a type I heterojunction.

What are the advantages of high switching frequency?

Power electronic circuits and RF applications require devices to work at high voltages and high frequencies. Low frequency devices will require larger passive components for a lesser efficient circuit. High frequency allows effective miniaturization of high voltage circuits allowing more portable appliances like EV chargers, motor control choppers, SMPS, etc. Let’s list some down:

• Smaller passive components like inductors, and capacitors: For a given energy, higher switching frequency (f_s) allows the use of smaller inductance (L) because energy transfer happens more often per second. The formula for current ripple in a buck converter is given below, where D = duty cycle. For the same ripple, increasing f_s allows smaller capacitance.

• Faster Dynamic Response: Higher fs​ means the converter can respond more quickly to load or input changes.
• Lower Filter Size: Higher frequency means noise and ripple shifted to higher spectrum, therefore it is easier to filter with smaller LC filters.
• Smaller Antennas: Length of antennas are half the wavelength of signal. Higher frequency gives lower wavelength.
• Higher Data Rates and Finer Resolution: At higher frequencies, more spectrum (bandwidth) is available resulting in higher data throughput. Resolution is inversely related to bandwidth.

Frequently Asked Questions (FAQ)

Ques: What is conduction band?

The conduction band is the range of electron energy levels in a solid where electrons are free to move through the crystal and contribute to electrical conduction. Electrons in the conduction band are no longer bound to individual atoms and can respond to an applied electric field.

Ques: What is fermi level?

The fermi energy level is the level where probability of finding electrons is ½.

Ques: What substrates are used in HEMT?

It depends on the heterostructure used. For GaAs HEMTs, GaAs substrates are typically employed because they provide excellent lattice matching with the epitaxial layers and support high electron mobility. In contrast, GaN HEMTs are commonly fabricated on silicon carbide (SiC), silicon (Si), sapphire (Al₂O₃), or native GaN substrates. SiC is preferred for high-power and high-frequency applications due to its superior thermal conductivity, while silicon offers a lower-cost alternative suitable for large-scale manufacturing. Native GaN substrates provide the highest crystal quality and lowest defect densities but are considerably more expensive and are available in very low wafer diameters, producing lower throughput.

Ques: Which type of scattering hampers HEMT performance the most?

The dominant scattering mechanism in a HEMT depends on the operating conditions. At room temperature and above, phonon scattering, particularly polar optical phonon scattering in GaN-based HEMTs, is generally the primary factor limiting electron mobility. While the formation of the 2DEG greatly suppresses ionized impurity scattering (by roughly an order of magnitude compared to bulk transport), phonon scattering is an intrinsic property of the crystal lattice and cannot be eliminated. As a result, once impurity and defect scattering are reduced, phonon scattering becomes the principal limitation on mobility and high-frequency device performance.